Low battery warning circuit

[@rt]

Hi
Is there a need for any more than a simple resistor divider in this situation?

I only have a digital input, and Vcc for the shift register is 5V.
In the real circuit the battery could be up to 13.8 Volts, so I realise the 150K resistor should be a higher value
to prevent the shift register pin ever seeing a voltage above it’s 5V supply rail.

This way I think diode clamping the shift register input can be avoided.

Software ensures this input must read the warning status 253 times consecutively before the status is actually acted on.

It has been suggested that I use the BC548 level shifter on this page, but I fail to see the point:
https://jeelabs.org/book/1504d/
The 12V and logic would be swapped to the other direction in this case.

[digsys]

Apart from the issues you mention - your circuit doesn't take into account -
Varying battery voltage - I'm assuming you are working directly off a battery ie 11.5V-ish to 13.8V-ish, so your reference will vary a lot.
Potentiometers drift a little on top of that.
Check out the TPS3700 - has 2 in-built UV and OV precision references, good to 18V, O/C output = ok for ANY logic output / input, cheap and easy.
I use them for all battery / voltage monitoring. Only require 2-3 resistors. The suggested circuits need no extra work.

[@rt]

I don’t quite understand.
The 5V output of the regulator that is powering the shift register is the stable voltage reference.
Therefore, I expect the high/low logic threshold voltage for its input to be relatively stable.
The battery voltage is supposed to vary.

I maybe see a problem if the battery was so low that the regulator couldn’t supply 5V, but the idea was to catch the condition before that happens.

[digsys]

OK a bit lost here. If you only need to know when the battery voltage drops to ~8.0V (assuming ~3V loss over the 5.0V regulators),
then that will work. I am assuming you are using a SLA or other type of Lead acid? in which case you need to protect that MORE than
just the logic side. That is why I suggested that part. SLAs should never go OV / UV. Is this correct?
At the very least, I'd put a 4v7 400mW zener on the wiper of the pot.

[@rt]

It will never be SLA. It’s 8xAA in series for nominal 12V,
So I realise the voltage will begin higher than 12V with fresh batteries (making the 150K resistor too low a value).
In the end this will be 3x18650 batteries (nominal 10.8V), so discharging those too far will be an issue, but that should just mean the 150K resistor needs to be adjusted yet again.
Understood about the zener.

[@rt]

The microcontroller reading the shift register pin can also turn power off BTW. This little drawing isn’t supposed to include that.
I just want a warning here giving the user a few minutes to save a GPS log, or to do it automatically while it still can.

[tautech]

You want a dedicated device for this and something you can tune for switch points and hysteresis.
Datasheet of ICL7665 attached for your perusal.
It's made by a couple of manufacturers and available in SMD or DIP.

[KL27x]

   Looks fine to me.

Software ensures this input must read the warning status 253 times consecutively before the status is actually acted on.

If you are using a micro, and have an extra pin, you could switch your ladder to only draw power during a reading.

253 times consecutively before the status is actually acted on.

One way I have done this for li ion battery is to have the device turn off automatically when a countdown timer reaches zero. And whenever the voltage is above the cutoff, this timer is reset. In my case, I made the timer about 1 minute and put in a visual LED indicator which kicked in at low voltage and pulsed faster and faster, reaching unity at the point it would automatically turn off. So if I would know a minute in advance when the device would shutoff because it would start blinking, and if I want to finish something, I could turn it off/on and continue using the device for another full minute. With alkalines, I guess you wouldn't need any auto shutoff.

[@rt]

I’m using that count as a timer to set Batlow status, then a warning on display, then that count is reset with another timer that requires at least 2 minutes between onscreen warnings. No extra pin here, just a spare shift register input. Circuit is fine so far, but it’s closer to the transistor circuit, and will draw a little current on its own.